STEREO MATCHING AS A NEAREST-NEIGHBOR PROBLEM

We propose a representation of images, called intrinsic curves, that t ransforms stereo matching from a search problem into a nearest-neighbo r problem. Intrinsic curves are the paths that a set of local image de scriptors trace as an image scanline is traversed from left to right. Intrinsic curves are ideally invariant with respect to disparity. Ster eo correspondence then becomes a trivial lookup problem in the ideal c ase. We also show how to use intrinsic curves to match real images in the presence of noise, brightness bias, contrast fluctuations, moderat e geometric distortion, image ambiguity, and occlusions. In this case, matching becomes a nearest-neighbor problem, even for very large disp arity values.